The Predictive Role of Metabolic Volume Segmentation Compared to Semiquantitative PET Parameters in Diagnosis of LVAD Infection using [18F]FDG Imaging

Abstract

PurposeLeft ventricular assisting device (LVAD) is a vital mechanical circulatory assist device for patients with end-stage heart disease, serving as either a bridge to transplantation or palliative destination therapy. Yet device infection represents a major lethal complication, warranting a multi-step, complex therapy approach including an urgent device exchange or heart transplantation. Still, timely diagnosis of site and extent of VAD-specific infection for a proper therapy planning poses challenges in regular clinical care. This single-center, retrospective study aimed to evaluate the impact of volumetric PET parameters with different thresholding compared to semiquantitative PET parameters for accurate diagnosis of VAD-specific infection.ProceduresSeventeen patients (1 female, 16 males; mean age 57 ± 11 years) underwent [18F]FDG imaging for suspected VAD-specific infection between April 2013 and October 2023. Various metabolic and volumetric PET parameters with different thresholding were collected for specific LVAD components including driveline entry point, subcutaneous driveline, pump pocket, inner cannula and outflow tract. Microbiology and clinical follow-up were used as the final diagnosis standard.ResultsNine of eleven patients with VAD-specific infection underwent urgent heart transplantation, and one had a surgical revision of LVAD. Two patients had non-VAD specific infections, and two had non-VAD related infections. Metabolic burden determination using a fixed absolute threshold provided the best outcome compared to relative thresholding or other metabolic SUV parameters. The total metabolic tumor volume (MTV) cutoff value was 9.3 cm3, and the corresponding sensitivity, specificity, accuracy, and AUC were 90.0%, 71.43%, 82.5%, and 0.814 (95% CI 0.555–0.958), respectively. The total lesion glycolysis (TLG) was 30.6, and the corresponding sensitivity, specificity, accuracy, and AUC were 90.0%, 71.4%, 82.5%, and 0.829 (95% CI 0.571–0.964), respectively.ConclusionsVolumetric PET parameters with fixed absolute thresholding appear to be a valuable auxiliary tool in the evaluation of [18F]FDG imaging to enhance the diagnostic accuracy of VAD-specific infection.